BLUE

Ket-arbon Emissions Joshi

@ketanjoshi.co

Fighting alongside my friends for the best possible pathway to climate justice and fossil fuel elimination. Anti-greenwashing. Creator / curator of Greensky: tinyurl.com/3jk8zrt8 ketanjoshi.co/ ketan.joshi85@gmail.com

13k followers1.9k following11.8k posts

KEketanjoshi.coOct 12, 2024 6:16pm

Something really notable - and this is consistent across the big tech orgs like Google and Microsoft etc - is that Meta never, ever cites generative AI when it comes to its "AI for climate" stuff

Open sourcing our model for sustainable concrete
Our climate-related AI initiatives, including Open Catalyst↗ and the Llama Impact Grants↗ program, are
advancing climate action for the broader community.
2024 Sustainability Report
Concrete currently accounts for over 8% of
carbon emissions worldwide.1
This can be
reduced by making adjustments to concrete
formulas that optimize for sustainability
performance and technical performance
criteria such as compressive strength gain
and workability. The formulation and testing
of concrete mixes to optimize for multiple
variables can be costly and time-consuming
without the use of AI.
To address this challenge, Meta developed an
open source model↗ that uses an innovative
approach to concrete formulation, integrating
multiple performance requirements in
collaboration with researchers at the
University of Illinois at Urbana-Champaign↗.
The model outputs were refined in partnership
with a local industry partner (Ozinga) to
generate concrete mixtures with up to a
40% lower carbon footprint than the regional
industry standard. This lower-carbon concrete
was used in multiple applications at our
DeKalb, IL, data center campus. Variations of
these mixtures have subsequently been used
in at least two other sites, and enabled Meta to
scale requirements to use low carbon concrete
across its entire portfolio for future designs in
specific applications.
We are partnering with top leaders in
sustainable construction to collaborate on the
expansion of this project, and utilize and fine
tune this model for thei

Forests harbor most of Earth’s terrestrial biodiversity and play a critical role in the uptake of carbon dioxide
from the atmosphere. Ecosystem services provided by forests underpin an essential defense against the
climate and biodiversity crises. However, critical gaps remain in the scientific understanding of the structure
and extent of global forests, particularly in dynamic and more dispersed forest systems such as agroforestry,
drylands forests and alpine forests, which together constitute more than a third of the world’s forests. Higher
resolution forest data has the potential to dramatically improve the monitoring of conservation and restoration
efforts globally. This is important for both forest-based carbon credit projects and other conservation efforts
focused on the preservation of forests.
We leveraged our DINOv2 AI training algorithm↗ and collaborated with World Resources Institute (WRI)↗ to
launch an open source map of global tree canopy height↗ at a 1-meter resolution, allowing the detection of
individual trees. We hope this high resolution canopy height data can ultimately be applied to forest biomass
and carbon stock monitoring. Remote sensing can make the monitoring and verification of carbon removal
projects more efficient and more accessible, especially for smaller scale pro

KJkevinjkircher.comOct 12, 2024 6:29pm

Such a sly move! We've been using data science and numerical optimization to improve design and operation of buildings, power grids, etc. for decades. Some people call this "AI," but it's basically unrelated to large language models.

Ket-arbon Emissions Joshi

@ketanjoshi.co

13k followers1.9k following11.8k posts